Working with HCPC and Class Specific Analysis
Andreas Roaldsnes
About this rpubs document - Heteropraxia and Homopraxia
The goal in this Rpubs document is to show how I combined variables from two different questions about parental practices to create a space of parental practices. The practices are gathered from a respondent (mother or father), who then answers questions about which organizations he or she either is active in (cultural organization, voluntary organization etc) or has previously been active in, and supplies the same for his or her spouse or cohabitor (what I term “The second adult in the household”). There are single parent households also, but these have been excluded for this particular analysis. Organizational history is then analysed together with audience consuption habits from the past four months, for first parent and for second parent (as reported by first parent). This results in a space of organized and audience activities.
The main goal is to look into how practices are structured, and especially look for gendered differences in roles in the household.
The academic use of this excercise is to analyse if there is a homologous relationship between parental practices and children’s practices, and secondly to investigate whether or not parental practices form mostly similar or dissimilar practices between parents (What Octobre 2008 terms Homopraxia and Heteropraxia), which she identifies as strengthening or weakening reproduction of spatial positions.
To do this I will
- Contruct data on organized activities
- Construct data on audience participation
- Combine data and run MCA
- Check if the logic structuring the space for men and women is the same using Class Specific Analysis
- Run HCPC-clustering to further analyse the dimensions of the MCA
- Use the clusters to explore Homopraxia and Heteropraxia
- … which could perhaps be used in my actual analysis of childrens space of organized activities and cultural transmission
Packages and data
library(tidyverse)
library(FactoMineR)
library(factoextra)
library(janitor)
library(corrplot)
library(haven)
library(kableExtra)
library(grid)
library(gridExtra)
# Henter inn omkodinger fra bfp runde1.
source("C:/Andreas/R/Rpubs/Source_RMD.R")
source_rmd("C:/Andreas/R/Rpubs/Grunnomkodinger bfp.Rmd") # Loads bfp and recodes it.
gdata::keep(bfp, sure = TRUE)
df <- read_sav("C:/Andreas/R/Rpubs/BFP/BKKN - kulturkonsum i barnefamilier - r1.sav")
df2 <- read_sav("C:/Andreas/R/Rpubs/BFP2/BKKN - kulturkonsum i barnefamilier - r2.sav")
df3 <- read_sav("C:/Andreas/R/Rpubs/BFP3/BKKN - kulturkonsum i barnefamilier - r3.sav")Constructing data for MCA
Participation in organizations for parents
praxis_f1 <- df %>%
filter(status == 1, r1reg2==2) %>%
haven::zap_label() %>% haven::zap_formats() %>% haven::zap_labels() %>%
select(ID_resp, voksen_kjonn, contains("r1q12_1_now"), contains("r1q12_1_prev")) %>%
mutate(`Cultural org` = ifelse(r1q12_1_now_1==1, "Active",
ifelse(r1q12_1_prev_1==1, "Previously active", "Never")),
`Religious org` = ifelse(r1q12_1_now_2==1, "Active",
ifelse(r1q12_1_prev_2==1, "Previously active", "Never")),
`Voluntary org` = ifelse(r1q12_1_now_3==1, "Active",
ifelse(r1q12_1_prev_3==1, "Previously active", "Never")),
`Political org` = ifelse(r1q12_1_now_4==1, "Active",
ifelse(r1q12_1_prev_4==1, "Previously active", "Never")),
`Athletics org` = ifelse(r1q12_1_now_5==1, "Active",
ifelse(r1q12_1_prev_5==1, "Previously active", "Never")),
`Outdoors org` = ifelse(r1q12_1_now_6==1, "Active",
ifelse(r1q12_1_prev_6==1, "Previously active", "Never")),
`Physical excercise` = ifelse(r1q12_1_now_7==1, "Active",
ifelse(r1q12_1_prev_7==1, "Previously active", "Never"))) %>%
mutate(FID = "F1") %>%
select(ID_resp, FID, voksen_kjonn, contains("org"), contains("exercise"))
praxis_f2 <- df %>%
filter(status == 1, r1reg2==2) %>%
haven::zap_label() %>% haven::zap_formats() %>% haven::zap_labels() %>%
select(ID_resp, voksen_kjonn, contains("r1q12_2_now"), contains("r1q12_2_prev")) %>%
mutate(`Cultural org` = ifelse(r1q12_2_now_1==1, "Active",
ifelse(r1q12_2_prev_1==1, "Previously active", "Never")),
`Religious org` = ifelse(r1q12_2_now_2==1, "Active",
ifelse(r1q12_2_prev_2==1, "Previously active", "Never")),
`Voluntary org` = ifelse(r1q12_2_now_3==1, "Active",
ifelse(r1q12_2_prev_3==1, "Previously active", "Never")),
`Political org` = ifelse(r1q12_2_now_4==1, "Active",
ifelse(r1q12_2_prev_4==1, "Previously active", "Never")),
`Athletics org` = ifelse(r1q12_2_now_5==1, "Active",
ifelse(r1q12_2_prev_5==1, "Previously active", "Never")),
`Outdoors org` = ifelse(r1q12_2_now_6==1, "Active",
ifelse(r1q12_2_prev_6==1, "Previously active", "Never")),
`Physical excercise` = ifelse(r1q12_2_now_7==1, "Active",
ifelse(r1q12_2_prev_7==1, "Previously active", "Never"))) %>%
mutate(FID = "F2") %>%
select(ID_resp, FID, voksen_kjonn, contains("org"), contains("exercise")) %>%
mutate(voksen_kjonn = ifelse(voksen_kjonn==2, 1, 2))
# Key is made from respondent-ID and FIC, which denotes if this row is
# a respondent (Called F1) or the second adult as reported by the respondent (called F2).
# When F1 is female, F2 is assumed to be male, and vice versa.
praxis <- bind_rows(praxis_f1, praxis_f2) %>%
mutate(Key = paste0(ID_resp, FID))
# Quick glance at initial MCA shows that the MCA struggles with
# Active and previously active, and that this dominates axis 1
# In the analysis.
# Recoding needed
res.praxis <- MCA(praxis %>% select(-c(ID_resp, voksen_kjonn, FID, Key)))
fviz_mca_var(res.praxis)
GDAtools::modif.rate(res.praxis)
# Recoding involves flattening of "Active" and "Previously active"
# Generating Yes and No
praxis_flat <- praxis %>%
mutate_at(vars(contains( names(praxis)[4:9] )), ~ifelse(. == "Never", "No", "Yes"))
# Generate a numerical version for use in paran:paran, a parallell analysis.
praxis_flat_num <- praxis %>%
mutate_at(vars(contains( names(praxis)[4:9] )), ~ifelse(. == "Never", 0, 1))Parrallel analysis with paran
Checking the data with a parallel analysis.
paran::paran(praxis_flat_num %>% select(-c(ID_resp, voksen_kjonn, FID, Key)))##
## Using eigendecomposition of correlation matrix.
## Computing: 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
##
##
## Results of Horn's Parallel Analysis for component retention
## 180 iterations, using the mean estimate
##
## --------------------------------------------------
## Component Adjusted Unadjusted Estimated
## Eigenvalue Eigenvalue Bias
## --------------------------------------------------
## 1 1.503497 1.540563 0.037066
## 2 1.005988 1.026642 0.020654
## --------------------------------------------------
##
## Adjusted eigenvalues > 1 indicate dimensions to retain.
## (2 components retained)Eigenvalues
GDAtools::modif.rate(res.praxis.flat)## $raw
## eigen rate cum.rate
## dim 1 0.2567606 25.67606 25.67606
## dim 2 0.1711071 17.11071 42.78677
## dim 3 0.1598462 15.98462 58.77139
## dim 4 0.1536742 15.36742 74.13881
## dim 5 0.1330831 13.30831 87.44712
## dim 6 0.1255288 12.55288 100.00000
##
## $modif
## mrate cum.mrate
## dim 1 99.7576683 99.75767
## dim 2 0.2423317 100.00000Visually checking the MCA
Contributions
| Var | Ctr.1 | Ctr.2 |
|---|---|---|
| Voluntary org_Yes | 23.4643676 | 0.0453990 |
| Cultural org_Yes | 18.7587424 | 3.1999372 |
| Religious org_Yes | 17.5662852 | 9.3975732 |
| Political org_Yes | 17.2294714 | 1.9594617 |
| Voluntary org_No | 6.4119764 | 0.0124059 |
| Outdoors org_Yes | 5.6538272 | 27.1969929 |
| Cultural org_No | 3.9448962 | 0.6729353 |
| Athletics org_Yes | 2.0977381 | 29.9996626 |
| Athletics org_No | 1.5717664 | 22.4777630 |
| Religious org_No | 1.5364959 | 0.8219913 |
| Political org_No | 0.9095216 | 0.1034375 |
| Outdoors org_No | 0.8549117 | 4.1124404 |
Audience participation data
##### Consumption ####
# First parent (F1)
consumpt_f1 <- df %>%
filter(status == 1, r1reg2==2) %>%
haven::zap_label() %>% haven::zap_formats() %>% haven::zap_labels() %>%
select(ID_resp, voksen_kjonn, contains("r1q17_1_")) %>%
mutate(FID = "F1")
# Gathering variable names from dataframe-labels.
la_b <- df %>%
select(ID_resp, contains("r1q17_1_")) %>%
labelled::var_label()
la <- tibble(Var = names(la_b), Values = unlist(la_b)) %>%
mutate(labels = gsub("\\s*\\([^\\)]+\\)","", Values))
la$labels[la$Var == "r1q17_1_1"] = "Classical-Jazz"
la$labels[la$Var == "r1q17_1_2"] = "Pop-etc"
names(consumpt_f1)[3:17] <- la$labels[2:16]
# Second adult in the household (F2)
consumpt_f2 <- df %>%
filter(status == 1, r1reg2==2) %>%
haven::zap_label() %>% haven::zap_formats() %>% haven::zap_labels() %>%
select(ID_resp, voksen_kjonn, contains("r1q17_2_")) %>%
mutate(voksen_kjonn = ifelse(voksen_kjonn==2, 1, 2),
FID = "F2")
la_b <- df %>%
select(ID_resp, contains("r1q17_2_")) %>%
labelled::var_label()
la <- tibble(Var = names(la_b), Values = unlist(la_b)) %>%
mutate(labels = gsub("\\s*\\([^\\)]+\\)","", Values))
la$labels[la$Var == "r1q17_2_1"] = "Classical-Jazz"
la$labels[la$Var == "r1q17_2_2"] = "Pop-etc"
names(consumpt_f2)[3:17] <- la$labels[2:16]
# Same Key as above
consumpt <- bind_rows(consumpt_f1, consumpt_f2) %>%
mutate(Key = paste0(ID_resp, FID))Frequencies of the categories
| Category | Share |
|---|---|
| Kino | 0.5145656 |
| Idretts- eller sportsarrangement | 0.4347770 |
| Folkefest eller annet offentlig utendørsarrangement | 0.3029131 |
| Låne bøker på bibliotek | 0.2958237 |
| Teater | 0.2626966 |
| Pop-etc | 0.2481310 |
| Museum eller kulturminne | 0.1992782 |
| Stand-up eller revy | 0.1469451 |
| Classical-Jazz | 0.1223253 |
| Kunstutstilling | 0.1161382 |
| Festival | 0.1113689 |
| Dans | 0.0766950 |
| Arrangement på bibliotek | 0.0653519 |
| Litteraturarrangement | 0.0540088 |
| Åpent verksted | 0.0339005 |
# Recode to character
consumpt_chr <- consumpt %>%
mutate_at(vars(contains( names(consumpt)[3:17] )), ~ifelse(. == 0, "No", "Yes"))# Cloud of individuals reveal something that resembles a comet tail.
# Needs some recoding
res_cons <- MCA(consumpt_chr %>% select(-c(ID_resp, voksen_kjonn, FID, Key)))
fviz_mca_var(res_cons)
fviz_mca_ind(res_cons, geom.ind = "point", alpha.ind = .2)
GDAtools::modif.rate(res_cons)
# Joining some categories to eliminate the comet tail.
cons_chr_rec <- consumpt_chr %>%
select(-(`Åpent verksted`)) %>%
mutate(Literature = ifelse(`Arrangement på bibliotek`=="Yes" | Litteraturarrangement == "Yes", "Yes", "No"),
Festival = ifelse(`Folkefest eller annet offentlig utendørsarrangement`=="Yes" | Festival == "Yes", "Yes", "No")) %>%
select(-c(`Arrangement på bibliotek`, Litteraturarrangement, `Låne bøker på bibliotek`, `Folkefest eller annet offentlig utendørsarrangement`))Parallel analysis with paran
# Paran
cons_num <- cons_chr_rec %>%
mutate_at(vars(contains( names(cons_chr_rec)[c(3:13,15)] )), ~ifelse(. == "Yes", 1, 0)) %>%
select(-c(ID_resp, voksen_kjonn, FID, Key))
paran::paran(cons_num)##
## Using eigendecomposition of correlation matrix.
## Computing: 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
##
##
## Results of Horn's Parallel Analysis for component retention
## 330 iterations, using the mean estimate
##
## --------------------------------------------------
## Component Adjusted Unadjusted Estimated
## Eigenvalue Eigenvalue Bias
## --------------------------------------------------
## 1 2.198644 2.258030 0.059386
## 2 1.195131 1.239192 0.044060
## --------------------------------------------------
##
## Adjusted eigenvalues > 1 indicate dimensions to retain.
## (2 components retained)MCA for audience participation
res_cons <- MCA(cons_chr_rec %>% select(-c(ID_resp, voksen_kjonn, FID, Key)))
res.cons.df <- inspect.mca(res_cons) %>%
mutate(fontfam = ifelse((Ctr.1>mean(Ctr.1) & Ctr.2>mean(Ctr.2)), 4,
ifelse(Ctr.1>mean(Ctr.1), 3,
ifelse(Ctr.2>mean(Ctr.2), 2, 1))))Eigenvalues
GDAtools::modif.rate(res_cons)## $raw
## eigen rate cum.rate
## dim 1 0.20527553 20.527553 20.52755
## dim 2 0.11265386 11.265386 31.79294
## dim 3 0.08931687 8.931687 40.72463
## dim 4 0.08754506 8.754506 49.47913
## dim 5 0.08117247 8.117247 57.59638
## dim 6 0.07779041 7.779041 65.37542
## dim 7 0.07724911 7.724911 73.10033
## dim 8 0.07436538 7.436538 80.53687
## dim 9 0.07131104 7.131104 87.66797
## dim 10 0.06950664 6.950664 94.61863
## dim 11 0.05381365 5.381365 100.00000
##
## $modif
## mrate cum.mrate
## dim 1 96.511092 96.51109
## dim 2 3.488908 100.00000Visualizing MCA for audience participation
fviz_mca_var(res_cons)
fviz_mca_ind(res_cons, geom.ind = "point", alpha.ind = .2)
Contributions, audience participation
| Var | Ctr.1 | Ctr.2 |
|---|---|---|
| Kunstutstilling_Yes | 16.7551793 | 4.4359820 |
| Museum eller kulturminne_Yes | 13.8342861 | 2.3051126 |
| Teater_Yes | 10.1724437 | 0.8091202 |
| Literature_Yes | 9.4301526 | 5.5346252 |
| Classical-Jazz_Yes | 8.1498980 | 4.6246517 |
| Pop-etc_Yes | 6.1987379 | 9.8151584 |
| Festival_Yes | 5.8650939 | 0.1093031 |
| Dans_Yes | 4.6415734 | 0.5483173 |
| Teater_No | 3.6243777 | 0.2882844 |
| Museum eller kulturminne_No | 3.4429823 | 0.5736806 |
| Kino_No | 3.2436547 | 7.6036403 |
| Festival_No | 3.1845080 | 0.0593471 |
| Kino_Yes | 3.0600209 | 7.1731737 |
| Kunstutstilling_No | 2.2016066 | 0.5828817 |
| Pop-etc_No | 2.0457004 | 3.2391874 |
| Stand-up eller revy_Yes | 1.1889509 | 25.2873112 |
| Classical-Jazz_No | 1.1358868 | 0.6445579 |
| Literature_No | 1.0510972 | 0.6168966 |
| Dans_No | 0.3855558 | 0.0455464 |
| Stand-up eller revy_No | 0.2048057 | 4.3559285 |
| Idretts- eller sportsarrangement_Yes | 0.1037116 | 12.0659814 |
| Idretts- eller sportsarrangement_No | 0.0797763 | 9.2813125 |
Data on volunteering added
## Done volunteer work recently #####
friv <- df %>% select(ID_resp, r1q30_1, r1q30_2)
friv_F1 <- friv %>%
select(ID_resp, r1q30_1) %>%
mutate(`Volunteered` = ifelse(r1q30_1==1, "Yes", "No")) %>%
select(-r1q30_1) %>%
mutate(FID = "F1")
friv_F2 <- friv %>%
select(ID_resp, r1q30_2) %>%
mutate(`Volunteered` = ifelse(r1q30_2==1, "Yes", "No")) %>%
select(-r1q30_2) %>%
mutate(FID = "F2")
friv.comb <- bind_rows(friv_F1, friv_F2) %>%
mutate(Key = paste0(ID_resp, FID))
MCA for the whole space
Combining the data sources
comb.all <- left_join(praxis_flat, cons_chr_rec, by = c("Key" = "Key")) %>%
left_join(friv.comb, by = c("Key" = "Key"))
comb.run <- select(comb.all, -c(contains("ID_resp"), contains("FID"), Key)) %>%
select(-voksen_kjonn.y) %>%
rename(VK = voksen_kjonn.x) %>%
mutate(VK = ifelse(VK==1, "Mann", "Kvinne"))Run MCA full space
res.comb <- MCA(comb.run, quali.sup = 1, graph=FALSE)Eigenvalues full space
GDAtools::modif.rate(res.comb)## $raw
## eigen rate cum.rate
## dim 1 0.14213286 14.213286 14.21329
## dim 2 0.08556556 8.556556 22.76984
## dim 3 0.07785418 7.785418 30.55526
## dim 4 0.05907273 5.907273 36.46253
## dim 5 0.05779534 5.779534 42.24207
## dim 6 0.05454356 5.454356 47.69642
## dim 7 0.05032264 5.032264 52.72869
## dim 8 0.04941781 4.941781 57.67047
## dim 9 0.04904059 4.904059 62.57453
## dim 10 0.04812520 4.812520 67.38705
## dim 11 0.04670440 4.670440 72.05749
## dim 12 0.04500707 4.500707 76.55819
## dim 13 0.04364425 4.364425 80.92262
## dim 14 0.04221457 4.221457 85.14407
## dim 15 0.04213663 4.213663 89.35774
## dim 16 0.03957265 3.957265 93.31500
## dim 17 0.03447042 3.447042 96.76204
## dim 18 0.03237955 3.237955 100.00000
##
## $modif
## mrate cum.mrate
## dim 1 84.11805014 84.11805
## dim 2 10.10678718 94.22484
## dim 3 5.58003891 99.80488
## dim 4 0.13882548 99.94370
## dim 5 0.05629829 100.00000Contributions and coordinates full space
res.comb.df <- inspect.mca(res.comb) %>%
mutate(fontfam = ifelse((Ctr.1>mean(Ctr.1) & Ctr.2>mean(Ctr.2)), 4,
ifelse(Ctr.1>mean(Ctr.1), 3,
ifelse(Ctr.2>mean(Ctr.2), 2, 1))))
res.comb.df %>%
select(Var, Ctr.1, Ctr.2, Ctr.3, Coord.1, Coord.2, Coord.3) %>%
arrange(desc(Ctr.1)) %>%
kbl() %>%
kable_classic()| Var | Ctr.1 | Ctr.2 | Ctr.3 | Coord.1 | Coord.2 | Coord.3 |
|---|---|---|---|---|---|---|
| Kunstutstilling_Yes | 12.0480409 | 4.7439841 | 0.0643384 | 1.6291242 | -0.7931765 | 0.0881099 |
| Museum eller kulturminne_Yes | 10.1614985 | 3.1488343 | 0.0064140 | 1.1421747 | -0.4933224 | 0.0212379 |
| Classical-Jazz_Yes | 7.1299544 | 1.1622668 | 1.9384619 | 1.2211490 | -0.3825432 | 0.4712460 |
| Teater_Yes | 7.0037984 | 2.8880600 | 1.8561317 | 0.8258917 | -0.4114922 | -0.3146693 |
| Literature_Yes | 6.9686589 | 3.9955703 | 0.7135870 | 1.3333470 | -0.7833579 | 0.3157808 |
| Voluntary org_Yes | 6.0909663 | 7.8661292 | 4.0645046 | 0.8521075 | 0.7513357 | 0.5151679 |
| Cultural org_Yes | 6.0128524 | 1.7457027 | 5.5677967 | 0.9409222 | 0.3933696 | 0.6701142 |
| Festival_Yes | 5.0123937 | 0.0499787 | 0.1246716 | 0.6036701 | -0.0467705 | -0.0704619 |
| Pop-etc_Yes | 4.1540321 | 0.4686538 | 8.1583109 | 0.6544522 | -0.1705577 | -0.6787922 |
| Political org_Yes | 3.6033348 | 1.8310683 | 5.4898536 | 1.3559257 | 0.7499600 | 1.2386759 |
| Dans_Yes | 3.3782516 | 1.1037537 | 0.0003674 | 1.0615633 | -0.4708019 | 0.0081936 |
| Festival_No | 2.7215264 | 0.0271364 | 0.0676916 | -0.3277684 | 0.0253945 | 0.0382580 |
| Volunteered_Yes | 2.5891097 | 17.8429078 | 0.0493458 | 0.4751683 | 0.9678484 | -0.0485502 |
| Museum eller kulturminne_No | 2.5289241 | 0.7836603 | 0.0015963 | -0.2842566 | 0.1227747 | -0.0052855 |
| Teater_No | 2.4954093 | 1.0289976 | 0.6613280 | -0.2942600 | 0.1466121 | 0.1121147 |
| Kino_No | 2.1534928 | 0.2124828 | 4.9521176 | -0.3368914 | 0.0821074 | 0.3781007 |
| Kino_Yes | 2.0315767 | 0.2004535 | 4.6717623 | 0.3178189 | -0.0774591 | -0.3566952 |
| Voluntary org_No | 1.6644443 | 2.1495331 | 1.1106844 | -0.2328506 | -0.2053133 | -0.1407770 |
| Kunstutstilling_No | 1.5830954 | 0.6233527 | 0.0084540 | -0.2140646 | 0.1042223 | -0.0115775 |
| Religious org_Yes | 1.3951445 | 5.9215238 | 12.3171591 | 0.6661556 | 1.0648418 | 1.4649240 |
| Pop-etc_No | 1.3709089 | 0.1546646 | 2.6923965 | -0.2159816 | 0.0562873 | 0.2240142 |
| Cultural org_No | 1.2644813 | 0.3671150 | 1.1708877 | -0.1978726 | -0.0827242 | -0.1409226 |
| Volunteered_No | 1.0749387 | 7.4079639 | 0.0204872 | -0.1972789 | -0.4018283 | 0.0201569 |
| Classical-Jazz_No | 0.9937328 | 0.1619902 | 0.2701719 | -0.1701969 | 0.0533167 | -0.0656796 |
| Athletics org_Yes | 0.7784618 | 7.8754484 | 5.4952313 | 0.2156313 | 0.5321488 | -0.4240137 |
| Literature_No | 0.7767359 | 0.4453515 | 0.0795373 | -0.1486166 | 0.0873141 | -0.0351973 |
| Stand-up eller revy_Yes | 0.6530794 | 0.0017922 | 13.8030116 | 0.3372011 | -0.0137055 | -1.1473254 |
| Athletics org_No | 0.5832759 | 5.9008151 | 4.1173965 | -0.1615655 | -0.3987217 | 0.3176996 |
| Outdoors org_Yes | 0.3839587 | 3.4532292 | 0.4046476 | 0.2734724 | 0.6363356 | 0.2077799 |
| Idretts- eller sportsarrangement_Yes | 0.3563742 | 8.5963748 | 9.2219531 | 0.1448116 | 0.5518361 | -0.5451994 |
| Dans_No | 0.2806170 | 0.0916841 | 0.0000305 | -0.0881796 | 0.0391075 | -0.0006806 |
| Idretts- eller sportsarrangement_No | 0.2741277 | 6.6124452 | 7.0936483 | -0.1113910 | -0.4244797 | 0.4193746 |
| Political org_No | 0.1902154 | 0.0966597 | 0.2898023 | -0.0715776 | -0.0395894 | -0.0653881 |
| Religious org_No | 0.1220311 | 0.5179466 | 1.0773630 | -0.0582676 | -0.0931401 | -0.1281346 |
| Stand-up eller revy_No | 0.1124978 | 0.0003087 | 2.3776720 | -0.0580854 | 0.0023609 | 0.1976354 |
| Outdoors org_No | 0.0580582 | 0.5221606 | 0.0611865 | -0.0413516 | -0.0962199 | -0.0314183 |
Visual inspection of MCA
fviz_mca_ind(res.comb, geom.ind = "point", alpha.ind = .1)
fviz_mca_var(res.comb)
My own visualizations
ggplot() +
geom_hline(yintercept = 0, linetype = "dashed", alpha = .5) +
geom_vline(xintercept = 0, linetype = "dashed", alpha = .5) +
geom_point(data = res.comb.df, aes(Coord.1, Coord.2), shape = 2) +
ggrepel::geom_text_repel(data = res.comb.df, aes(Coord.1, Coord.2, label = Var, fontface = fontfam), max.overlaps = Inf, min.segment.length = 0, seed = 42, box.padding = 0.5) +
labs(title = "Axis 1 and 2") +
coord_fixed()
ggplot() +
geom_hline(yintercept = 0, linetype = "dashed", alpha = .5) +
geom_vline(xintercept = 0, linetype = "dashed", alpha = .5) +
geom_point(data = res.comb.df, aes(Coord.2, Coord.3), shape = 2) +
ggrepel::geom_text_repel(data = res.comb.df, aes(Coord.2, Coord.3, label = Var, fontface = fontfam), max.overlaps = Inf, min.segment.length = 0, seed = 42, box.padding = 0.5) +
labs(title = "Axis 2 and 3") +
coord_fixed()
HCPC for the space
res.hcpc <- HCPC(res.comb, nb.clust = -1)Cluster visualization
fviz_cluster(res.hcpc, geom = "point")
Tabyl of clusters
res.hcpc$data.clust$clust %>% tabyl() %>% kbl() %>% kable_classic()| . | n | percent |
|---|---|---|
| 1 | 3792 | 0.4887858 |
| 2 | 2518 | 0.3245682 |
| 3 | 1448 | 0.1866460 |
Cluster tables
cl1 <- make.clust(res.hcpc, 1)
cl2 <- make.clust(res.hcpc, 2)
cl3 <- make.clust(res.hcpc, 3)Cluster 1 - “Less active”
The first and largest cluster is mainly characterized in the lack of volunteering, and a smaller investment into legitime culture. One simplification could be that while cluster 2 attends sports events and cluster 3 attend highly legitimate arts events, cluster 1 rarely attend either of these. As with the difference in attendance, cluster 1 less frequently see a history of organizational participation, and very little volunteering.
Cluster 1 is in fact not only underrepresented in highly legitimate events and sports event, it is underrepresented in all event types, and all organizational types. The category with greates distance from cluster 1 to global is Arts exhibitions-yes.
Men are ever so slightly overrepresented, with 51.3 to 48.6.
It is tempting to label cluster 1 “Less active”, but there is also a greater aversion towards highly legitimate events.
| Variable | Cla/Mod | Mod/Cla | Global | v.test | p.value |
|---|---|---|---|---|---|
| Volunteered=Volunteered_No | 64.155 | 92.748 | 70.663 | Inf | 0.000 |
| Idretts-.eller.sportsarrangement=Idretts- eller sportsarrangement_No | 71.152 | 82.278 | 56.522 | Inf | 0.000 |
| Athletics.org=Athletics org_No | 68.568 | 80.195 | 57.167 | Inf | 0.000 |
| Volunteered=Volunteered_Yes | 12.083 | 7.252 | 29.337 | -Inf | 0.000 |
| Idretts-.eller.sportsarrangement=Idretts- eller sportsarrangement_Yes | 19.923 | 17.722 | 43.478 | -Inf | 0.000 |
| Athletics.org=Athletics org_Yes | 22.600 | 19.805 | 42.833 | -Inf | 0.000 |
| Kunstutstilling=Kunstutstilling_No | 54.003 | 97.653 | 88.386 | 26.588 | 0.000 |
| Kunstutstilling=Kunstutstilling_Yes | 9.878 | 2.347 | 11.614 | -26.588 | 0.000 |
| Museum.eller.kulturminne=Museum eller kulturminne_No | 55.618 | 91.113 | 80.072 | 24.415 | 0.000 |
| Museum.eller.kulturminne=Museum eller kulturminne_Yes | 21.798 | 8.887 | 19.928 | -24.415 | 0.000 |
| Voluntary.org=Voluntary org_No | 55.818 | 89.689 | 78.538 | 23.905 | 0.000 |
| Voluntary.org=Voluntary org_Yes | 23.483 | 10.311 | 21.462 | -23.905 | 0.000 |
| Pop-etc=Pop-etc_No | 56.129 | 86.340 | 75.187 | 22.586 | 0.000 |
| Pop-etc=Pop-etc_Yes | 26.909 | 13.660 | 24.813 | -22.586 | 0.000 |
| Festival=Festival_No | 57.220 | 75.870 | 64.811 | 20.084 | 0.000 |
| Festival=Festival_Yes | 33.516 | 24.130 | 35.189 | -20.084 | 0.000 |
| Kino=Kino_No | 60.409 | 59.995 | 48.543 | 19.811 | 0.000 |
| Kino=Kino_Yes | 38.001 | 40.005 | 51.457 | -19.811 | 0.000 |
| Teater=Teater_No | 55.507 | 83.729 | 73.730 | 19.781 | 0.000 |
| Teater=Teater_Yes | 30.275 | 16.271 | 26.270 | -19.781 | 0.000 |
| Cultural.org=Cultural org_No | 53.526 | 90.480 | 82.624 | 18.161 | 0.000 |
| Cultural.org=Cultural org_Yes | 26.780 | 9.520 | 17.376 | -18.161 | 0.000 |
| Classical-Jazz=Classical-Jazz_No | 52.475 | 94.225 | 87.767 | 17.396 | 0.000 |
| Classical-Jazz=Classical-Jazz_Yes | 23.077 | 5.775 | 12.233 | -17.396 | 0.000 |
| Literature=Literature_No | 51.633 | 95.042 | 89.972 | 14.875 | 0.000 |
| Literature=Literature_Yes | 24.165 | 4.958 | 10.028 | -14.875 | 0.000 |
| Stand-up.eller.revy=Stand-up eller revy_No | 52.312 | 91.297 | 85.305 | 14.783 | 0.000 |
| Stand-up.eller.revy=Stand-up eller revy_Yes | 28.947 | 8.703 | 14.695 | -14.783 | 0.000 |
| Dans=Dans_No | 50.845 | 96.044 | 92.330 | 12.277 | 0.000 |
| Dans=Dans_Yes | 25.210 | 3.956 | 7.670 | -12.277 | 0.000 |
| Political.org=Political org_No | 50.156 | 97.468 | 94.986 | 10.018 | 0.000 |
| Political.org=Political org_Yes | 24.679 | 2.532 | 5.014 | -10.018 | 0.000 |
| Outdoors.org=Outdoors org_No | 50.185 | 89.188 | 86.865 | 5.938 | 0.000 |
| Outdoors.org=Outdoors org_Yes | 40.236 | 10.812 | 13.135 | -5.938 | 0.000 |
| Religious.org=Religious org_No | 49.678 | 93.460 | 91.957 | 4.774 | 0.000 |
| Religious.org=Religious org_Yes | 39.744 | 6.540 | 8.043 | -4.774 | 0.000 |
| VK=Mann | 50.193 | 51.345 | 50.000 | 2.316 | 0.021 |
| VK=Kvinne | 47.564 | 48.655 | 50.000 | -2.316 | 0.021 |
Cluster 2 - Sports and fun
Cluster 2 is a cluster oriented towards sports. The cluster is twice as likely to have attended a sports event, and almost twice as likely to have a history of involvement in an sports or athletics organization (Athletics org). The cluster does’nt only have this history and participation structure, they are also very likely to volunteer, with 56 % versus the global 29 %.
In addition, there is an aversion towards traditional high culture expressions, with low attendence in categories Literature, Arts exhibitions, Museums, Classical and jazz concerts, but they are not inactive in the participation realm. Cluster 2 has above average attendence in Stand-up performances, and pop (electonica, metal, rock etc) concerts. These are all categories which have lower traditional legitimacy (even more purely oriented towards Having fun). Cluster 2 is closer to average in the categories Theatre and Dance.
Cluster 2 is only slighter lower than average when it comes to a history of participation in a cultural organization, but half the average for history in a political organization.
It is tempting to label the cluster Sports and fun.
| Variable | Cla/Mod | Mod/Cla | Global | v.test | p.value |
|---|---|---|---|---|---|
| Idretts-.eller.sportsarrangement=Idretts- eller sportsarrangement_Yes | 62.467 | 83.678 | 43.478 | Inf | 0 |
| Athletics.org=Athletics org_Yes | 57.478 | 75.854 | 42.833 | Inf | 0 |
| Idretts-.eller.sportsarrangement=Idretts- eller sportsarrangement_No | 9.373 | 16.322 | 56.522 | -Inf | 0 |
| Athletics.org=Athletics org_No | 13.709 | 24.146 | 57.167 | -Inf | 0 |
| Volunteered=Volunteered_Yes | 62.522 | 56.513 | 29.337 | 35.847 | 0 |
| Volunteered=Volunteered_No | 19.974 | 43.487 | 70.663 | -35.847 | 0 |
| Kunstutstilling=Kunstutstilling_No | 35.774 | 97.419 | 88.386 | 19.188 | 0 |
| Kunstutstilling=Kunstutstilling_Yes | 7.214 | 2.581 | 11.614 | -19.188 | 0 |
| Literature=Literature_No | 35.158 | 97.458 | 89.972 | 16.854 | 0 |
| Literature=Literature_Yes | 8.226 | 2.542 | 10.028 | -16.854 | 0 |
| Museum.eller.kulturminne=Museum eller kulturminne_No | 36.140 | 89.158 | 80.072 | 14.473 | 0 |
| Museum.eller.kulturminne=Museum eller kulturminne_Yes | 17.658 | 10.842 | 19.928 | -14.473 | 0 |
| Stand-up.eller.revy=Stand-up eller revy_Yes | 50.000 | 22.637 | 14.695 | 13.329 | 0 |
| Stand-up.eller.revy=Stand-up eller revy_No | 29.435 | 77.363 | 85.305 | -13.329 | 0 |
| Classical-Jazz=Classical-Jazz_No | 34.880 | 94.321 | 87.767 | 12.940 | 0 |
| Classical-Jazz=Classical-Jazz_Yes | 15.068 | 5.679 | 12.233 | -12.940 | 0 |
| Political.org=Political org_No | 33.370 | 97.657 | 94.986 | 7.941 | 0 |
| Political.org=Political org_Yes | 15.167 | 2.343 | 5.014 | -7.941 | 0 |
| Pop-etc=Pop-etc_Yes | 39.429 | 30.143 | 24.813 | 7.454 | 0 |
| Pop-etc=Pop-etc_No | 30.156 | 69.857 | 75.187 | -7.454 | 0 |
| Kino=Kino_Yes | 36.298 | 57.546 | 51.457 | 7.448 | 0 |
| Kino=Kino_No | 28.386 | 42.454 | 48.543 | -7.448 | 0 |
| VK=Mann | 36.324 | 55.957 | 50.000 | 7.279 | 0 |
| VK=Kvinne | 28.590 | 44.043 | 50.000 | -7.279 | 0 |
| Teater=Teater_No | 34.720 | 78.872 | 73.730 | 7.221 | 0 |
| Teater=Teater_Yes | 26.104 | 21.128 | 26.270 | -7.221 | 0 |
| Cultural.org=Cultural org_No | 33.682 | 85.743 | 82.624 | 5.090 | 0 |
| Cultural.org=Cultural org_Yes | 26.632 | 14.257 | 17.376 | -5.090 | 0 |
| Dans=Dans_No | 33.184 | 94.400 | 92.330 | 4.862 | 0 |
| Dans=Dans_Yes | 23.697 | 5.600 | 7.670 | -4.862 | 0 |
| Voluntary.org=Voluntary org_Yes | 36.937 | 24.424 | 21.462 | 4.371 | 0 |
| Voluntary.org=Voluntary org_No | 31.233 | 75.576 | 78.538 | -4.371 | 0 |
| Outdoors.org=Outdoors org_Yes | 37.684 | 15.250 | 13.135 | 3.784 | 0 |
| Outdoors.org=Outdoors org_No | 31.666 | 84.750 | 86.865 | -3.784 | 0 |
| Religious.org=Religious org_No | 33.011 | 93.527 | 91.957 | 3.578 | 0 |
| Religious.org=Religious org_Yes | 26.122 | 6.473 | 8.043 | -3.578 | 0 |
Cluster 3 - Very active and culturally oriented
Cluster 3 is five times more likely to have participated in an Art exhibition, three times more likely to have attended a Museum, a literature event and a classical or jazz concert or a theatre event, and doble the likelihood for a Dance event and a Festival event.
Cluster 3 is culturally engaged, and especially engaged with traditionally legitimate forms of culture. Cluster 3 have twice the likelihood of having a history of participation in a cultural organization.
There are no special aversions, as participation in less legitimate art forms are at average, even sports events.
Women are overrepresented in the cluster by 63 % vs 36 % men.
It is tempting to label the cluster Very active and culturally oriented.
| Variable | Cla/Mod | Mod/Cla | Global | v.test | p.value |
|---|---|---|---|---|---|
| Museum.eller.kulturminne=Museum eller kulturminne_Yes | 60.543 | 64.641 | 19.928 | Inf | 0.000 |
| Kunstutstilling=Kunstutstilling_Yes | 82.908 | 51.588 | 11.614 | Inf | 0.000 |
| Museum.eller.kulturminne=Museum eller kulturminne_No | 8.242 | 35.359 | 80.072 | -Inf | 0.000 |
| Kunstutstilling=Kunstutstilling_No | 10.223 | 48.412 | 88.386 | -Inf | 0.000 |
| Literature=Literature_Yes | 67.609 | 36.326 | 10.028 | 32.218 | 0.000 |
| Literature=Literature_No | 13.209 | 63.674 | 89.972 | -32.218 | 0.000 |
| Classical-Jazz=Classical-Jazz_Yes | 61.855 | 40.539 | 12.233 | 32.193 | 0.000 |
| Classical-Jazz=Classical-Jazz_No | 12.645 | 59.461 | 87.767 | -32.193 | 0.000 |
| Teater=Teater_Yes | 43.621 | 61.395 | 26.270 | 31.843 | 0.000 |
| Teater=Teater_No | 9.773 | 38.605 | 73.730 | -31.843 | 0.000 |
| Cultural.org=Cultural org_Yes | 46.588 | 43.370 | 17.376 | 26.506 | 0.000 |
| Cultural.org=Cultural org_No | 12.793 | 56.630 | 82.624 | -26.506 | 0.000 |
| Festival=Festival_Yes | 34.212 | 64.503 | 35.189 | 25.339 | 0.000 |
| Festival=Festival_No | 10.223 | 35.497 | 64.811 | -25.339 | 0.000 |
| Voluntary.org=Voluntary org_Yes | 39.580 | 45.511 | 21.462 | 23.146 | 0.000 |
| Voluntary.org=Voluntary org_No | 12.949 | 54.489 | 78.538 | -23.146 | 0.000 |
| Dans=Dans_Yes | 51.092 | 20.994 | 7.670 | 18.746 | 0.000 |
| Dans=Dans_No | 15.971 | 79.006 | 92.330 | -18.746 | 0.000 |
| Political.org=Political org_Yes | 60.154 | 16.160 | 5.014 | 18.706 | 0.000 |
| Political.org=Political org_No | 16.474 | 83.840 | 94.986 | -18.706 | 0.000 |
| Pop-etc=Pop-etc_Yes | 33.662 | 44.751 | 24.813 | 18.578 | 0.000 |
| Pop-etc=Pop-etc_No | 13.715 | 55.249 | 75.187 | -18.578 | 0.000 |
| Kino=Kino_Yes | 25.701 | 70.856 | 51.457 | 16.602 | 0.000 |
| Kino=Kino_No | 11.206 | 29.144 | 48.543 | -16.602 | 0.000 |
| VK=Kvinne | 23.846 | 63.881 | 50.000 | 11.772 | 0.000 |
| VK=Mann | 13.483 | 36.119 | 50.000 | -11.772 | 0.000 |
| Religious.org=Religious org_Yes | 34.135 | 14.710 | 8.043 | 9.616 | 0.000 |
| Religious.org=Religious org_No | 17.311 | 85.290 | 91.957 | -9.616 | 0.000 |
| Volunteered=Volunteered_Yes | 25.395 | 39.917 | 29.337 | 9.590 | 0.000 |
| Volunteered=Volunteered_No | 15.870 | 60.083 | 70.663 | -9.590 | 0.000 |
| Outdoors.org=Outdoors org_Yes | 22.080 | 15.539 | 13.135 | 2.951 | 0.003 |
| Outdoors.org=Outdoors org_No | 18.148 | 84.461 | 86.865 | -2.951 | 0.003 |
| Athletics.org=Athletics org_Yes | 19.922 | 45.718 | 42.833 | 2.455 | 0.014 |
| Athletics.org=Athletics org_No | 17.723 | 54.282 | 57.167 | -2.455 | 0.014 |
| Stand-up.eller.revy=Stand-up eller revy_Yes | 21.053 | 16.575 | 14.695 | 2.216 | 0.027 |
| Stand-up.eller.revy=Stand-up eller revy_No | 18.253 | 83.425 | 85.305 | -2.216 | 0.027 |
| Idretts-.eller.sportsarrangement=Idretts- eller sportsarrangement_No | 19.475 | 58.978 | 56.522 | 2.092 | 0.036 |
| Idretts-.eller.sportsarrangement=Idretts- eller sportsarrangement_Yes | 17.610 | 41.022 | 43.478 | -2.092 | 0.036 |
CSA
Create a MCA with package soc.ca
# SOC MCA
soc.data <- comb.all
soc.active <- soc.data[, c(4:9, 13:22, 24, 26)]
soc.sups <- soc.data[, c(1:3, 10:12, 23, 25, 27)]
soc.sups <- select(soc.sups, voksen_kjonn.x, FID) %>% rename(VK = voksen_kjonn.x)
soc.result <- soc.ca::soc.mca(soc.active)Print result of MCA
Same as with factominer.
soc.result## Specific Multiple Correspondence Analysis:
##
## Statistics Scree plot
## Active dimensions: 5 | 1. 84.1% ******************************************
## Dimensions explaining 80% of inertia: 1 | 2. 10.1% *****
## Active modalities: 36 | 3. 5.6% ***
## Supplementary modalities: 0 | 4. 0.1%
## Individuals: 7758 | 5. 0.1%
## Share of passive mass: 0 | 6. 0.0%
## Number of passive modalities: 0 | 7. 0.0%
##
## The 18 active variables: [No. modalities - share of variance]
##
## Cultural org [2 - 6%] Religious org [2 - 6%]
## Voluntary org [2 - 6%] Political org [2 - 6%]
## Athletics org [2 - 6%] Outdoors org [2 - 6%]
## Classical-Jazz [2 - 6%] Pop-etc [2 - 6%]
## Teater [2 - 6%] Dans [2 - 6%]
## Stand-up eller revy [2 - 6%] Kino [2 - 6%]
## Kunstutstilling [2 - 6%] Museum eller kulturminne [2 - 6%]
## Festival [2 - 6%] Idretts- eller sportsarrangement [2 - 6%]
## Literature [2 - 6%] Volunteered [2 - 6%]CSA on gender
Is the logic of the space the same for men and for women?
# CSA på kjønn
class.kjonn <- which(soc.sups$VK == "1")
csa.res.menn <- soc.ca::soc.csa(soc.result, class.kjonn)
class.kjonn <- which(soc.sups$VK == "2")
csa.res.kvinner <- soc.ca::soc.csa(soc.result, class.kjonn)Cosine similarity
Cosine similarity is very high for men. The axes are close to 100 percent in similarity.
Cosine similarity for the Women CSA show that axes 1 and 3 are very much the same as in the global MCA, but that axis 2 is structured a little bit differently for women. While similarity between axes 2 is 95 % for men, it is 87 % for women. Still, this is not very much.
Next, lets visualize the modalities between the CSAs and the global space.
##
## Measures for Class Specific Multiple Correspondence Analysis:
##
##
## Cosine similarity:
##
## MCA: 1 MCA: 2 MCA: 3 MCA: 4 MCA: 5
## CSA: 1 0.988319406 -0.084060127 -0.061454473 -0.057024618 -0.035779008
## CSA: 2 0.028603388 0.874333816 -0.478996477 -0.035579411 0.149052633
## CSA: 3 0.049666198 0.430700449 0.865567760 -0.000252663 0.153994351
## CSA: 4 -0.011945528 -0.105744061 -0.013199737 -0.631904154 0.475312143
## CSA: 5 0.013829052 -0.134472777 -0.017541201 0.418722315 0.698378232
## MCA: 1&2 MCA: 1&3 MCA: 1&4 MCA: 1&5 MCA: 2&3
## CSA: 1 0.991887772 0.990228206 0.989963158 0.988966827 0.104128561
## CSA: 2 0.874801564 0.479849746 0.045651378 0.151772333 0.996943954
## CSA: 3 0.433554619 0.866991510 0.049666841 0.161805413 0.966804232
## CSA: 4 0.106416644 0.017802491 0.632017053 0.475462227 0.106564720
## CSA: 5 0.135181990 0.022336884 0.418950618 0.698515138 0.135612025
## MCA: 2&4 MCA: 2&5 MCA: 3&4 MCA: 3&5 MCA: 4&5
## CSA: 1 0.101577123 0.091357771 0.083835907 0.071111108 0.067319718
## CSA: 2 0.875057436 0.886947749 0.480316062 0.501651585 0.153240275
## CSA: 3 0.430700523 0.457402598 0.865567796 0.879159716 0.153994559
## CSA: 4 0.640690772 0.486932685 0.632042002 0.475495391 0.790711384
## CSA: 5 0.439785522 0.711206779 0.419089574 0.698598489 0.814285289
## MCA: 1&2&3 MCA: 2&3&4 MCA: 3&4&5
## CSA: 1 0.993789719 0.118720530 0.091151504
## CSA: 2 0.997354201 0.997578640 0.502911729
## CSA: 3 0.968079105 0.966804265 0.879159753
## CSA: 4 0.107232155 0.640826731 0.790821551
## CSA: 5 0.136315311 0.440135205 0.814474202
##
##
##
## Cosine angles:
##
## MCA: 1 MCA: 2 MCA: 3 MCA: 4 MCA: 5 MCA: 1&2
## CSA: 1 8.8 85.2 86.5 86.7 87.9 7.3
## CSA: 2 88.4 29.0 61.4 88.0 81.4 29.0
## CSA: 3 87.2 64.5 30.1 90.0 81.1 64.3
## CSA: 4 89.3 83.9 89.2 50.8 61.6 83.9
## CSA: 5 89.2 82.3 89.0 65.2 45.7 82.2
## MCA: 1&3 MCA: 1&4 MCA: 1&5 MCA: 2&3 MCA: 2&4 MCA: 2&5
## CSA: 1 8.0 8.1 8.5 84.0 84.2 84.8
## CSA: 2 61.3 87.4 81.3 4.5 28.9 27.5
## CSA: 3 29.9 87.2 80.7 14.8 64.5 62.8
## CSA: 4 89.0 50.8 61.6 83.9 50.2 60.9
## CSA: 5 88.7 65.2 45.7 82.2 63.9 44.7
## MCA: 3&4 MCA: 3&5 MCA: 4&5 MCA: 1&2&3 MCA: 2&3&4 MCA: 3&4&5
## CSA: 1 85.2 85.9 86.1 6.4 83.2 84.8
## CSA: 2 61.3 59.9 81.2 4.2 4.0 59.8
## CSA: 3 30.1 28.5 81.1 14.5 14.8 28.5
## CSA: 4 50.8 61.6 37.7 83.8 50.1 37.7
## CSA: 5 65.2 45.7 35.5 82.2 63.9 35.5
##
## ##
## Measures for Class Specific Multiple Correspondence Analysis:
##
##
## Cosine similarity:
##
## MCA: 1 MCA: 2 MCA: 3 MCA: 4 MCA: 5
## CSA: 1 0.967775238 0.234598229 0.141429796 0.111596073 -0.017697155
## CSA: 2 -0.161835666 0.951034517 0.136116489 -0.057491133 -0.117010829
## CSA: 3 -0.051505596 -0.130609505 0.957798094 -0.185250888 -0.009345486
## CSA: 4 -0.046255709 -0.027440169 0.152959324 0.825879607 0.041791809
## CSA: 5 -0.032375348 0.062823175 0.062336626 0.200481993 0.836281528
## MCA: 1&2 MCA: 1&3 MCA: 1&4 MCA: 1&5 MCA: 2&3
## CSA: 1 0.995803816 0.978054855 0.974188172 0.967937034 0.273931956
## CSA: 2 0.964705881 0.211467448 0.171744034 0.199705576 0.960725950
## CSA: 3 0.140398252 0.959181951 0.192277711 0.052346580 0.966662316
## CSA: 4 0.053782465 0.159800329 0.827173933 0.062338960 0.155401150
## CSA: 5 0.070674709 0.070242566 0.203079277 0.836907974 0.088502013
## MCA: 2&4 MCA: 2&5 MCA: 3&4 MCA: 3&5 MCA: 4&5
## CSA: 1 0.259788399 0.235264783 0.180155685 0.142532721 0.112990587
## CSA: 2 0.952770635 0.958205712 0.147759700 0.179497167 0.130371640
## CSA: 3 0.226664365 0.130943426 0.975548605 0.957843686 0.185486468
## CSA: 4 0.826335337 0.049995181 0.839924806 0.158565790 0.826936322
## CSA: 5 0.210094695 0.838637911 0.209949719 0.838601604 0.859976642
## MCA: 1&2&3 MCA: 2&3&4 MCA: 3&4&5
## CSA: 1 1.005797011 0.295791142 0.181022816
## CSA: 2 0.974261328 0.962444587 0.188479344
## CSA: 3 0.968033501 0.984252978 0.975593368
## CSA: 4 0.162139163 0.840372919 0.840963873
## CSA: 5 0.094237835 0.219147521 0.862232961
##
##
##
## Cosine angles:
##
## MCA: 1 MCA: 2 MCA: 3 MCA: 4 MCA: 5 MCA: 1&2
## CSA: 1 14.6 76.4 81.9 83.6 89.0 5.3
## CSA: 2 80.7 18.0 82.2 86.7 83.3 15.3
## CSA: 3 87.0 82.5 16.7 79.3 89.5 81.9
## CSA: 4 87.3 88.4 81.2 34.3 87.6 86.9
## CSA: 5 88.1 86.4 86.4 78.4 33.3 85.9
## MCA: 1&3 MCA: 1&4 MCA: 1&5 MCA: 2&3 MCA: 2&4 MCA: 2&5
## CSA: 1 12.0 13.0 14.5 74.1 74.9 76.4
## CSA: 2 77.8 80.1 78.5 16.1 17.7 16.6
## CSA: 3 16.4 78.9 87.0 14.8 76.9 82.5
## CSA: 4 80.8 34.2 86.4 81.1 34.3 87.1
## CSA: 5 86.0 78.3 33.2 84.9 77.9 33.0
## MCA: 3&4 MCA: 3&5 MCA: 4&5 MCA: 1&2&3 MCA: 2&3&4 MCA: 3&4&5
## CSA: 1 79.6 81.8 83.5 NaN 72.8 79.6
## CSA: 2 81.5 79.7 82.5 13.0 15.8 79.1
## CSA: 3 12.7 16.7 79.3 14.5 10.2 12.7
## CSA: 4 32.9 80.9 34.2 80.7 32.8 32.8
## CSA: 5 77.9 33.0 30.7 84.6 77.3 30.4
##
## a <- soc.ca::map.active(csa.res.menn) + labs(title = "CSA, menn", subtitle = "Map of active modalities")
b <- soc.ca::map.active(csa.res.kvinner) + labs(title = "CSA, kvinner", subtitle = "Map of active modalities")
c <- soc.ca::map.active(soc.result) + labs(title = "MCA - hele rommet", subtitle = "Map of active modalities")a
b
c
Comparison between the male and female space
Having trouble interpreting comparisons between soc.ca::map.active on the CSA’s. The following maps the CSA’s in the same space, using color and geom_path to draw lines showing changes in the space by gender.
inspect.csa <- function(csa.object){
temp.df <- tibble(Var = csa.object[["names.mod"]],
VarQ = csa.object[["variable"]],
Dim.1 = csa.object[["coord.mod"]][,1],
Dim.2 = csa.object[["coord.mod"]][,2],
Dim.3 = csa.object[["coord.mod"]][,3],
Ctr.1 = csa.object[["ctr.mod"]][,1],
Ctr.2 = csa.object[["ctr.mod"]][,2],
Ctr.3 = csa.object[["ctr.mod"]][,3],
Cor.1 = csa.object[["cor.mod"]][,1],
Cor.2 = csa.object[["cor.mod"]][,2],
Cor.3 = csa.object[["cor.mod"]][,3])
}
ins.csa <- inspect.csa(csa.res.kvinner) %>%
mutate(fontfam = ifelse((Ctr.1>mean(Ctr.1) & Ctr.2>mean(Ctr.2)), 4,
ifelse(Ctr.1>mean(Ctr.1), 3,
ifelse(Ctr.2>mean(Ctr.2), 2, 1))),
alpvar = ifelse(fontfam==1, .5, 1),
sizevar = ifelse(fontfam==1, 1, 2))
ins.csa.menn <- inspect.csa(csa.res.menn) %>%
mutate(fontfam = ifelse((Ctr.1>mean(Ctr.1) & Ctr.2>mean(Ctr.2)), 4,
ifelse(Ctr.1>mean(Ctr.1), 3,
ifelse(Ctr.2>mean(Ctr.2), 2, 1))),
alpvar = ifelse(fontfam==1, .5, 1),
sizevar = ifelse(fontfam==1, 1, 2))
csa.comp <- bind_rows(ins.csa %>% mutate(CSA = "Kvinner"),
ins.csa.menn %>% mutate(CSA = "Menn")) %>%
filter(fontfam!=1)
csa.comp.lj <- left_join(ins.csa %>% mutate(CSA = "Kvinner"),
ins.csa.menn %>% mutate(CSA = "Menn"),
by = c("Var" = "Var"))
### Main space
# ggplot() +
# geom_hline(yintercept = 0, linetype = "dashed", alpha = .5) +
# geom_vline(xintercept = 0, linetype = "dashed", alpha = .5) +
# geom_point(data = ins.csa, aes(Dim.1, Dim.2, alpha = alpvar), shape = 2) +
# ggrepel::geom_text_repel(data = ins.csa, aes(Dim.1, Dim.2, label = Var, fontface = fontfam, alpha = alpvar, size = sizevar),
# max.overlaps = Inf, min.segment.length = 0, seed = 42, box.padding = 0.5) +
# coord_fixed() +
# scale_alpha_continuous(range = c(.4,1)) +
# scale_size_continuous(range = c(3,4)) +
# theme_minimal() +
# theme(legend.position = "none")
iterate.ggp <- ggplot()
for (u in unique(csa.comp$Var)){
iterate.ggp <- iterate.ggp + geom_path(data = csa.comp[csa.comp$Var==u, ], aes(Dim.1, Dim.2), linetype = "dotted", size = 1)
}
iterate.ggp +
geom_hline(yintercept = 0, linetype = "dashed", alpha = .5) +
geom_vline(xintercept = 0, linetype = "dashed", alpha = .5) +
geom_point(data = csa.comp, aes(Dim.1, Dim.2, alpha = alpvar, color = CSA, shape = as.character(fontfam+10), size = ifelse(Ctr.1>Ctr.2, Ctr.1, Ctr.2))) +
ggrepel::geom_text_repel(data = csa.comp, aes(Dim.1, Dim.2, label = Var, fontface = fontfam, alpha = alpvar),
max.overlaps = Inf, min.segment.length = 0, seed = 42, box.padding = 0.5) +
scale_size_continuous(range = c(2,6)) +
scale_alpha_continuous(range = c(.7, 1)) +
coord_fixed() +
scale_shape_manual(name = "Contributions",
labels = c("Axis 1", "Axis 2", "Both axes"),
values = c(15, 16, 17)) +
scale_color_manual(name = "CSA Type",
labels = c("CSA:Women", "CSA:Men"),
values = c("red", "blue")) +
guides(shape = guide_legend(override.aes = list(size = 5)),
size = "none",
alpha = "none")
# Array
soc.ca::map.csa.mca.array(csa.res.menn, ndim = 3)
soc.ca::map.csa.mca.array(csa.res.kvinner, ndim = 3)
# mindre like i andre og tredjeaksen. Kvinnene avviker litt, men ikke mye. Det ser ut som om menn er mer orientert mot organisasjoner.
# soc.ca::map.csa.mca(csa.res, mca.dim = 1, csa.dim = 1)
# soc.ca::map.csa.mca(csa.res, mca.dim = 2, csa.dim = 2)
# soc.ca::csa.measures(csa.res.kvinner)
# soc.ca::csa.measures(csa.res.menn)
# Cosine similarity er svært nært 1 for alle tre dimensjonene. Svært små forskjeller på menn og kvinner. (rommets logikk)Using CSA to validate
# Hva om vi kun bruker respondenten som har svart for seg selv?
# Bare F1
# Det er det samme rommet. Det ser ut som om kvinnene har svart godt for mennene sine, og omvendt.
# Greit å ha en statistisk god indikasjon på.
f1.soc <- bind_cols(soc.active, soc.sups)
f1.soc <- filter(f1.soc, FID == "F1")
f1.active <- f1.soc[, c(1:18)]
f1.sup <- f1.soc[, c(20)]
f1.res <- soc.ca::soc.mca(f1.active, f1.sup)
class.kjonn <- which(f1.sup$VK == "1")
f1.csa.res.menn <- soc.ca::soc.csa(f1.res, class.kjonn)
class.kjonn <- which(f1.sup$VK == "2")
f1.csa.res.kvinner <- soc.ca::soc.csa(f1.res, class.kjonn)
soc.ca::csa.measures(f1.csa.res.kvinner)
soc.ca::csa.measures(f1.csa.res.menn)Homopraxia and Heteropraxia
Making new variables to look at if the practices differ strongly between parents.
# Bind clusters from hcpc to original dataframe for the MCA, along with the ID_resp, the ID of respondents.
cl.binded <- bind_cols(tibble(Cluster = res.hcpc[["data.clust"]][["clust"]]),
comb.all %>% select(ID_resp),
comb.run)
# Making a for loop.
# For every couple in the dataframe (recall that each
# ID_resp has answered key questionnarie items for their partner)
# And within each partner-couple, we check if they belong to the
# same cluster of practices.
empty.list = list() #To store result from for loop
indices <- 1 # Counter to place each result in the previously empty list.
for (resp in unique(cl.binded$ID_resp)){
temp.df <- cl.binded[cl.binded$ID_resp==resp, ]
temp.df$NumCl <- as.numeric(temp.df$Cluster) # Save a numerical version of the clusters
little.df <- tibble(ID_resp = resp,
Praxia = ifelse(temp.df$Cluster[1]==temp.df$Cluster[2], "Homopraxia", "Heteropraxia"),
ClusterComp = paste0(temp.df$Cluster[1], ":", temp.df$Cluster[2]),
F_A = paste0(temp.df$VK[1],":",temp.df$Cluster[1]),
F_B = paste0(temp.df$VK[2],":",temp.df$Cluster[2]))
empty.list[[indices]] <- little.df
indices <- indices + 1
}
praxia <- data.table::rbindlist(empty.list) # Nifty little function from data.table package
praxia$ClusterComp2 <- praxia$ClusterComp
praxia$ClusterComp2[praxia$ClusterComp2=="1:2"] <- "2:1"
praxia$ClusterComp2[praxia$ClusterComp2=="1:3"] <- "3:1"
praxia$ClusterComp2[praxia$ClusterComp2=="2:3"] <- "3:2"
# Make some refined variables with the results.
praxia <- praxia %>%
mutate(
PraxiaComposite =
ifelse(Praxia=="Homopraxia" & ClusterComp2 == "3:3", "Homopraxia Cultured",
ifelse(Praxia=="Homopraxia" & ClusterComp2 == "2:2", "Homopraxia Sports",
ifelse(Praxia=="Homopraxia" & ClusterComp2 == "1:1", "Homopraxia Less active",
ifelse(Praxia!="Homopraxia" & ClusterComp2 == "2:1", "Heteropraxia Sports-Less active",
ifelse(Praxia!="Homopraxia" & ClusterComp2 == "3:1", "Heteropraxia Cultured-Less active",
ifelse(Praxia!="Homopraxia" & ClusterComp2 == "3:2", "Heteropraxia Cultured-Sports", "Mistake"))))))) %>%
mutate(Mor = ifelse(grepl("Kvinne", F_A), F_A, F_B),
Far = ifelse(grepl("Mann", F_A), F_A, F_B)) %>%
mutate(Mother = ifelse(grepl("1", Mor), "Less active",
ifelse(grepl("2", Mor), "Sports", "Cultured")),
Father = ifelse(grepl("1", Far), "Less active",
ifelse(grepl("2", Far), "Sports", "Cultured")))
Homopraxia most common
The majority of parent couples are in the homopraxia category, meaning they occupy the same cluster of practices.
The largest category is the homopraxis of less active. It’s biggest in the detailed Praxia variable at 34 percent. Homopraxia Cultured, Heteropraxia Sports-Cultured and Heteropraxia Cultured-Less active are almost equal in size at 10, 9 and 8 percent. These are good conditions to examine the differential structuring of these parental practices on children’s practices.
| praxia$PraxiaComposite | n | percent |
|---|---|---|
| Heteropraxia Cultured-Less active | 353 | 0.0910028 |
| Heteropraxia Cultured-Sports | 305 | 0.0786285 |
| Heteropraxia Sports-Less active | 785 | 0.2023717 |
| Homopraxia Cultured | 395 | 0.1018304 |
| Homopraxia Less active | 1327 | 0.3420985 |
| Homopraxia Sports | 714 | 0.1840681 |
| praxia$Praxia | n | percent |
|---|---|---|
| Heteropraxia | 1443 | 0.3720031 |
| Homopraxia | 2436 | 0.6279969 |
## `summarise()` has grouped output by 'Mother'. You can override using the `.groups` argument.| Mother | Father | N | Frequency |
|---|---|---|---|
| Cultured | Cultured | 395 | 0.1018304 |
| Cultured | Less active | 277 | 0.0714102 |
| Cultured | Sports | 253 | 0.0652230 |
| Less active | Cultured | 76 | 0.0195927 |
| Less active | Less active | 1327 | 0.3420985 |
| Less active | Sports | 442 | 0.1139469 |
| Sports | Cultured | 52 | 0.0134055 |
| Sports | Less active | 343 | 0.0884249 |
| Sports | Sports | 714 | 0.1840681 |
Part II - presenting children’s space
Now to the point. Am I able to use clusters from parent’s space to predict children’s belonging in cluster’s in the space of organized lifestyles for children?
Run analysis of the Children’s space of lifestyles in background (no echo), and displaying modalities and cloud of individuals.
Cloud of modalities - Childrens’s organized lifestyles
Cloud of individuals - Children’s organized lifestyles
Does parent’s practices explain children’s practices?
Clusters from the above MCA on the Space of Organized Lifestyles for Children.
##
## Call:
## glm(formula = Dep_var ~ Kjonn + PC, family = "binomial", data = prax)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6362 -0.6189 -0.4405 -0.3963 2.2729
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.50462 0.15895 -15.758 < 2e-16 ***
## KjonnJente 0.88695 0.12596 7.041 1.90e-12 ***
## PCHeteropraxia Sports-Less active 0.06224 0.19935 0.312 0.755
## PCHomopraxia Sports 0.22056 0.19435 1.135 0.256
## PCHeteropraxia Cultured-Less active 2.30876 0.21874 10.555 < 2e-16 ***
## PCHomopraxia Cultured 2.65212 0.20362 13.025 < 2e-16 ***
## PCHeteropraxia Cultured-Sports 1.68654 0.21206 7.953 1.82e-15 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 2088.1 on 1884 degrees of freedom
## Residual deviance: 1693.3 on 1878 degrees of freedom
## (10 observations deleted due to missingness)
## AIC: 1707.3
##
## Number of Fisher Scoring iterations: 4## Joining, by = "RID"
##
## Call:
## glm(formula = Dep_var ~ Kjonn + Mother * Father, family = "binomial",
## data = prax)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6357 -0.6646 -0.4407 -0.3785 2.3116
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.50336 0.15895 -15.750 < 2e-16 ***
## KjonnJente 0.88507 0.12604 7.022 2.18e-12 ***
## MotherSports 0.23695 0.24695 0.960 0.337
## MotherCultured 2.31970 0.23480 9.880 < 2e-16 ***
## FatherSports -0.09671 0.24906 -0.388 0.698
## FatherCultured 2.26300 0.40607 5.573 2.51e-08 ***
## MotherSports:FatherSports 0.08031 0.35292 0.228 0.820
## MotherCultured:FatherSports -0.52779 0.36071 -1.463 0.143
## MotherSports:FatherCultured -0.86152 0.61434 -1.402 0.161
## MotherCultured:FatherCultured -1.93097 0.47313 -4.081 4.48e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 2088.1 on 1884 degrees of freedom
## Residual deviance: 1692.0 on 1875 degrees of freedom
## (10 observations deleted due to missingness)
## AIC: 1712
##
## Number of Fisher Scoring iterations: 4## Joining, by = "RID"##
## Call:
## glm(formula = Dep_var ~ Kjonn * Mother + Kjonn * Father, family = "binomial",
## data = prax)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.7477 -0.6229 -0.4368 -0.3562 2.3619
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.30176 0.19116 -12.041 < 2e-16 ***
## KjonnJente 0.76055 0.23861 3.187 0.00144 **
## MotherSports 0.29670 0.27220 1.090 0.27572
## MotherCultured 1.91615 0.25223 7.597 3.03e-14 ***
## FatherSports -0.42409 0.23151 -1.832 0.06697 .
## FatherCultured 0.73474 0.26877 2.734 0.00626 **
## KjonnJente:MotherSports -0.07304 0.33887 -0.216 0.82935
## KjonnJente:MotherCultured -0.19651 0.32778 -0.600 0.54883
## KjonnJente:FatherSports 0.36760 0.29429 1.249 0.21162
## KjonnJente:FatherCultured 0.36919 0.37184 0.993 0.32077
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 2088.1 on 1884 degrees of freedom
## Residual deviance: 1708.1 on 1875 degrees of freedom
## (10 observations deleted due to missingness)
## AIC: 1728.1
##
## Number of Fisher Scoring iterations: 5
## Mother = Less active:
## contrast odds.ratio SE df null z.ratio p.value
## Sports / Less active 0.908 0.226 Inf 1 -0.388 0.6978
## Cultured / Less active 9.612 3.903 Inf 1 5.573 <.0001
## Cultured / Sports 10.588 4.621 Inf 1 5.407 <.0001
##
## Mother = Sports:
## contrast odds.ratio SE df null z.ratio p.value
## Sports / Less active 0.984 0.246 Inf 1 -0.066 0.9477
## Cultured / Less active 4.061 1.877 Inf 1 3.033 0.0024
## Cultured / Sports 4.128 1.801 Inf 1 3.249 0.0012
##
## Mother = Cultured:
## contrast odds.ratio SE df null z.ratio p.value
## Sports / Less active 0.536 0.140 Inf 1 -2.395 0.0166
## Cultured / Less active 1.394 0.339 Inf 1 1.366 0.1720
## Cultured / Sports 2.603 0.606 Inf 1 4.107 <.0001
##
## Results are averaged over the levels of: Kjonn
## Tests are performed on the log odds ratio scale
